Improving the housing-unit method for small-area population estimation using remote-sensing and GIS information

Small-area population estimates for a non-census year are essential for supporting a wide variety of planning processes. Many demographic or geographic-information-based models have been developed for generating small-area population estimates. Little research, however, attempted to integrate these two types of models to achieve a better estimation. This study explores the feasibility of incorporating geographic information system (GIS), remote-sensing and demographic data into the housing- unit (HU)method,a popular dem ographicmodel,toestimatesmall-area population in Grafton, WI, USA. In particular, two major components of the HU method, HU counts and persons per household (PPH), are obtained by modelling their relationships with demographic and geographic factors using a sequence of ordinary least- squares (OLS) regression models. Analysis of results indicates that spatial factors derived from remote sensing and GIS datasets, together with demographic information, can significantly improve the accuracy of small-area population estimation.